BCG vaccination in children

More than 100 million doses of BCG vaccine are used each year worldwide. The role of this vaccine in the global control of tuberculosis has generated controversy for several decades and has often been underestimated. Although it has not curtailed the global epidemic, BCG immunisation in children has undoubtedly reduced childhood morbidity and mortality from tuberculosis. In the linked randomised controlled trial (doi:10.1136/bmj.a2052), Hawkridge and colleagues compare the effectiveness of intradermal versus percutaneous BCG vaccination in infants from birth to 2 years of age.1

Four high quality trials have shown that BCG is about 70-80% effective in preventing disease when given to mycobacteria naive newborns.2 Efficacy was lower in trials of older children and adults, but it now seems that such trials probably included subjects who had been exposed to mycobacteria despite having negative baseline skin tests (newer in vitro assays are more sensitive than skin tests in detecting previous infection with Mycobacterium tuberculosis or non-tuberculous mycobacteria3). Children or adults who have been exposed to mycobacteria are either already protected against tuberculosis or are unable to sustain the BCG replication needed for protection.4 Thus, the efficacy of BCG will be absent or low, and trials including these subjects should not be considered when assessing the efficacy of routine childhood immunisation with BCG.

Variations in efficacy have been proposed for different strains of BCG, with reduced efficacy seen for strains with greater numbers of serial passages.5 However, because almost all species of natural or vaccine induced mycobacteria that have been studied (BCG, M microti, non-tuberculous mycobacteria, M tuberculosis) seem to confer cross protection against the subsequent mycobacterial disease that has been studied (tuberculosis, leprosy, and disease from Mycobacterium ulcerans or Mycobacterium avium),2 differences in the efficacy of BCG are probably caused by differences in trial design rather than small genetic differences between substrains of BCG.6

Different BCG strains and different methods of BCG production have different patterns of local reactogenicity.7 Serious side effects, such as disseminated BCG, can probably occur with all strains and have been well documented. Where feasible, HIV testing is recommended before giving BCG to newborns. However, because such testing is not widely available in resource poor settings, routine use of BCG should not be curtailed until HIV testing is possible.8

Another important question is whether the method of giving BCG—intradermal injection or percutaneous multiple puncture—affects efficacy or side effects. Immunogenicity studies show that the intradermal route results in a higher frequency of scarring, skin test conversion, and in vitro immunogenicity.9 However, the efficacy of the two different methods has never been compared in a large trial.

In their study performed in South Africa, Hawkridge and colleagues found no significant difference in rates of tuberculosis in the first 2 years of life between intradermal or percutaneous administration of Japanese 172 BCG.1 Although immediate systemic symptoms, duration of drainage, and minor local reactions were not studied routinely, the authors found no significant differences in the incidence of the uncommon and more serious reactions for which BCG recipients sought medical care—lymphadenitis, keloids, and disseminated BCG. The authors comment that neither route of administration worked “particularly well,” but this conclusion cannot be drawn in a study that had no placebo group.

Opinions differ on which method of administration is more practical and consistent. Intradermal injection needs considerable training to be carried out correctly, but percutaneous administration often results in inadequate delivery of vaccine because not all tines of the applicator may penetrate the skin.10 It is difficult to know whether routine percutaneous administration would be as effective as was reported by Hawkridge and colleagues.1 Because no validated surrogate of clinical protection is available, no test can show that immunisation has been effective.

The South African study had the virtue of providing explicit case definitions for paediatric tuberculosis to facilitate interpretation of the findings. Two modifications of the definitions would be useful in subsequent studies. Firstly, because this study was done in an endemic setting, where laboratories were handling many positive specimens and where cross contamination would have been possible, it would have been advisable to have a more rigorous definition of definite tuberculosis that required multiple colonies of M tuberculosis from a single specimen or multiple positive sputum specimens.11 Secondly, although the authors’ diagnostic algorithm is suitable for initial decisions about treatment, a research definition of clinical “probable” or “possible” tuberculosis should include criteria on the presence or absence of a radiological or clinical response to treatment.11 12 Nevertheless, the absence of a significant difference in the rate of the most rigorously defined end point, definite tuberculosis, supports the non-inferiority of the percutaneous method in this particular research setting.

Should national policies be influenced by the present study? If health workers can be appropriately trained, the intradermal route remains the most reliable and safe method of giving BCG. But percutaneous administration can now be considered an effective alternative when other considerations make it preferable, although high quality applicators and standard methodology are necessary.